InFlowOp: Label-Free In-Flow Multi-Agent Workflow Optimization Gains up to +11.97%
Xuehang Guo · hf · 2026-10-02
Problem
Multi-agent workflows built on LLMs must decide task granularity, agent assignment, and when to create specialists up front — success is unknown until execution, and fixing faults usually needs reference answers, graded outcomes, or full re-runs.
Method
- InFlowOp prices every decision with one label-free cost balancing agent competence vs. runtime expense; decomposition granularity and assignment follow from this cost bidirectionally instead of fixed templates.
- During execution, faults are corrected with the cheapest move using the same cost function — no re-execution or retraining.
Results
- Introduces Braid, a benchmark requiring multi-agent coordination beyond single-agent capability.
- Outperforms single-agent baselines by up to +11.97%, with +9.64% from in-flow optimization across domains and backbones.
More from coding & agent
- Dev says Claude and Claude Code are getting worse day by day — Pavan_Belagatti · 2026-10-02
- Dev hooks Reels into Claude Code so he can doomscroll while the agent codes — yoimnotkesku · 2026-10-02
- Book a flight in 2 min or endure a 10-min AI agent call? Devs argue agent UX — jxnlco · 2026-10-02
- How llama.cpp Finds Safe Seams: Inside the Model-Loading Compatibility Hub — Mahmoud_Zalt · 2026-10-02
- Argo-Bench Pits Data Agents Against a 7.5B-Row Warehouse; Best Model Clears Only 34.8% of Tasks — textql · 2026-10-02
- Flexport launches MCP letting AI tools ship containers worldwide, customs included — ycombinator · 2026-10-02